Overview
Anonymization settings let you control how sensitive information is masked in conversation transcripts. Each language has its own anonymization configuration, including whether anonymization is active, a default masking interval, agent channel detection, and lists of sensitive words to redact before and after processing.
This is useful for protecting personally identifiable information (PII), payment details, or other confidential data that appears in calls, chats, and emails processed by the platform.
How It Works
Accessing Anonymization Settings
Open Settings from the main navigation, then select System. On the System page, select the Anonymization tab. This opens the Anonymizer Settings page, which lists languages and their current anonymization configuration.
Who Can Use Anonymization Settings
Only users with the system management permission can view or edit anonymization settings. Users without this permission cannot access the Anonymization section.
Viewing Language Settings
The settings page displays a table with the following columns for each language:
Name — the language name.
Active — whether anonymization is enabled for this language.
Default interval — the anonymization interval in seconds.
Detect agent channel — whether the system detects and treats the agent's audio channel differently.
Pre-sensitive words — words redacted before transcript processing.
Post-sensitive words — words redacted after transcript processing.
Filtered and All Languages Modes
By default, the list shows only the languages relevant to your organization's language configuration. This filtered view keeps the list focused on languages that your system actually uses for recognition.
To see the full catalog of all available languages, toggle All languages in the top-right corner of the page. This is useful when you need to configure anonymization for a language not shown in the default filtered view.
Bilingual Language Expansion
When your organization uses a bilingual or related language configuration, the filtered view automatically expands to show all languages that may be active during recognition:
Russian configurations also show Ukrainian, because bilingual recognition may produce transcripts in either language.
Ukrainian configurations also show Russian, for the same reason.
Bilingual configurations show both Russian and Ukrainian alongside the bilingual entry.
This ensures you can configure anonymization for every language the system may actually use, even when the primary language setting is a single bilingual code.
Editing a Language's Anonymization Settings
Click the edit icon (pencil) on the language row you want to configure.
On the edit form, adjust the available settings:
Active — enable or disable anonymization for this language.
Default anonymization interval — the number of seconds of audio to mask around detected sensitive content. Defaults to 5 seconds.
Detect agent channel — when enabled, the system identifies which audio channel belongs to the agent, which can affect how anonymization is applied. Enabled by default.
Pre-sensitive words (one per line) — words to redact before the main transcript processing. Enter one word per line.
Post-sensitive words (one per line) — words to redact after the main transcript processing. Enter one word per line.
Click Save to apply changes.
After saving, you are returned to the language list and a confirmation message appears.
Important Notes
Each language is configured independently. Enabling anonymization for one language does not affect others.
The language name on the edit form is read-only — you are editing the settings for that specific language, not renaming it.
Empty lines and extra whitespace in sensitive word lists are ignored automatically.
Changes take effect for new transcript processing. Previously processed transcripts are not retroactively re-anonymized.